Storage and eruption of silicic magma across the transition from dominantly effusive to caldera-forming states at an arc volcano (Santorini, Greece)

• Main Authors: Druitt, Timothy H., Fabbro, Gareth N., Costa, Fidel
• Other Authors: Earth Observatory of Singapore
• Format: Article
• Language: English
• Published: 2019
• Subjects: Calderas; Crystal Zoning; DRNTU::Social sciences::Geography
• Online Access: https://hdl.handle.net/10356/104831; http://hdl.handle.net/10220/47898
• Institution: Nanyang Technological University

Caldera-forming eruptions at Santorini discharge large volumes of silicic magma from upper crustal reservoirs. Sequences of smaller eruptions preceding the main explosive eruptions can provide insight into the conditions of the plumbing system that lead to caldera-forming events, which is important for interpreting monitoring data. We analysed textures, zoning patterns, trace element compositions, and crystal residence timescales calculated at pre-eruptive conditions (near-liquidus residence timescales; NLRT) of plagioclase and orthopyroxene phenocrysts from two eruptive units on Santorini: an ∼2 km3 sequence of dacitic lavas erupted between 39 and 25 ka (the Therasia dome complex), and a caldera-forming dacitic eruption that occurred no more than 2800 ± 1400 years after the last Therasia lava (the 21·8 ka, >10 km3 Cape Riva eruption). The study builds on our previous work, which showed that the Therasia and Cape Riva dacites, although similar in most major elements, differ in some trace element contents and were derived from different source magmas and crystal mushes. Contents of K and La in plagioclase phenocrysts mirror those in the respective host magmas (higher in Therasia, lower in Cape Riva), showing that plagioclase provenance can be determined using these elements. Very few plagioclase cores from the Cape Riva dacite were recycled from Therasia magmas; the majority were derived from lower-K, lower-La magmas and mushes related to the Cape Riva eruption itself. Despite the very different magma volumes and eruptive fluxes, plagioclase and orthopyroxene crystals from the two dacite series have remarkably similar textures and major element compositions. Furthermore, Mg diffusion profiles in plagioclase and Mg–Fe diffusion profiles in orthopyroxene yield similar ranges of NLRT, most ranging from years to centuries. Some orthopyroxene crystals exhibit Al sector zoning indicative of rapid growth. Processes driving crystallization appear to have been similar in the two systems, despite the differences of scale. Based on previously published phase diagrams and melt inclusion volatile barometry for the Cape Riva dacite, we infer that in each case crystallization of plagioclase rims and orthopyroxene took place centuries to years prior to eruption owing to volatile-saturated decompression (± cooling) as the dacitic melts (plus entrained plagioclase antecrystic cores) ascended from the middle crust (10–16 km) into the upper crust (4–6 km), where they resided until eruption a few years to decades later. Between 39 and 25 ka, multiple small volumes of Therasia-type dacitic magma were emplaced in the upper crust, where they either froze or were subsequently erupted. From about 25 ka onwards, large volumes of Cape Riva-type dacitic magma, sourced from a different mid-crustal reservoir, began to ascend into the upper crust. Runaway drainage of this magma source, peaking during the decades to years prior to the Cape Riva eruption, led to establishment of a well-mixed magma chamber in the upper crust that was discharged during the caldera-forming Cape Riva event.